You are on page 1of 7



Con J. Doolan, Danielle J. Moreau and Laura A. Brooks
School of Mechanical Engineering, The University of Adelaide, Adelaide, SA 5005, Australia

The aerodynamic noise production mechanisms of modern horizontal axis wind turbines are reviewed. An engineering
analysis of the time and frequency scales from three noise sources, leading edge turbulence interaction noise, trailing edge
noise and blade-tower interaction noise is presented. The analysis shows that noise sources are present from low-frequencies
(1-4 Hz) to over 500 Hz for a representative wind turbine. The results of the analysis are used to explain amplitude modulation
observed during noise measurements at a European wind farm. Daytime noise measurements close to a South Australian
wind farm are also presented that show amplitude modulation. The paper concludes with a description of conceptual ideas
for the control of wind turbine noise.

INTRODUCTION Gearboxes on modern turbines are now very quiet [4] and
Climate change policies have forced governments therefore the dominant noise sources are located on the blade.
around the world to mandate large increases in wind power. These noise sources are aeroacoustic in origin and in order
Consequently, wind power is now one the fastest growing to understand them, a review of blade aerodynamics is first
energy sources, with worldwide generation predicted to necessary.
increase from 150 TWh in 2008 to 1068 TWh (per annum) by Figure 1 shows an idealised picture of a wind turbine
2030 [1]. In Australia, wind energy production is set to increase outer blade tip moving through air. The major aerodynamic
from 4 TWh in 2007-8 to over 40 TWh by 2030. phenomena that influence noise are shown. Ahead of the blade
Wind energy increases will mean that many more wind is atmospheric (or other) turbulence. When the blade interacts
turbines will be installed, inevitably closer to more people with these turbulent eddies, unsteady lift is generated by the
and their residences. Noise from wind turbines is a serious blade. The unsteady lift creates a dipole-like sound source
and controversial issue and it can be expected to become more located at the blade leading edge [5]. This is called inflow or
of a concern as wind power production is increased. Surveys leading-edge interaction noise and has a dipole-like directivity
[2] show that noise from wind turbines is annoying to people pattern.
and that it is perceived to be more annoying than other forms The flow of air over the blade surface creates a boundary
of industrial noise at the same level. To accommodate the layer, due to the viscous shear present between the blade and
expected increase in the number of installed wind farms and the air. The flow conditions on large wind turbine means this
to reduce public disquiet, there needs to be more research and boundary layer will usually transition to a turbulent state by
development into how wind turbine noise is generated and how the time the air reaches the trailing edge. Turbulence by itself
it can be controlled. is a very inefficient radiator of sound [6], but when turbulent
The purpose of this paper is to review the aeroacoustic eddies pass a sharp edge (such as the trailing edge of a wind
source mechanisms that are on a wind turbine blade and turbine blade), the acoustic waves created by turbulence are
possible methods for reducing their strengths. An engineering reinforced via an edge diffraction mechanism [7], making
analysis is performed that gives an indication of the frequencies them much more efficient. This is known as trailing edge noise
that contain most of the energy for each type of source. Some [8] and is the major noise source on a wind turbine [4, 9, 10].
recently published results on wind farm noise will be discussed An important quality of trailing edge noise is its directivity
that suggest that the noise from multiple wind turbines can pattern, which is different from a monopole or dipole. Figure 2
interact, creating intermittent regions of increased noise illustrates the directivity pattern of trailing edge noise, assuming
amplitude. Daytime noise measurements taken several hundred that the frequency of sound emitted from the trailing edge is high
meters from a South Australian wind farm are also presented. enough so that the airfoil can be considered a semi-infinite half-
These measurements show noticeable amplitude modulation plane. Most of the sound is radiated forward of the blade (in
that is similar to that of European data. An explanation for the what is known as a cardioid directivity pattern), in the direction
noise phenomena is suggested in this paper along with some of rotation, while little is radiated behind. This explains the
conceptual ideas for its control. “swish” character of wind turbine noise whereby an observer
on the ground will periodically receive fluctuations in acoustic
WIND TURBINE AERODYNAMIC NOISE energy as the blade rotates. Here, “swish” is defined as the
GENERATION MECHANISMS amplitude modulation of broadband aerodynamic noise created
by the blades at the blade passing frequency, which is usually
The major noise sources on a wind turbine are located at
about 1 Hz [11]. The received acoustic signal has both a high
the gearbox and the fast moving outer blade tip region [3].

Acoustics Australia Vol. 40, No. 1, April 2012 - 7

frequency broadband character (due to turbulence in the blade generation is similar to trailing edge noise as it involves the
boundary layer) and a low frequency amplitude modulation (due interaction of turbulence with an edge. It is not believed to be
to the combination of the directivity function and convective as significant as the trailing edge source [4]; however, more
amplification of sound due to blade rotation). It is not clear work needs to be done in this area.
whether reports of “thumping” noise [12] at large distances are The other noise source to be considered is airfoil tonal
due to swish or another effect such as blade tower interaction. noise [13]. Here, discrete vortices form either in the boundary
The interaction of the rotor blade with the tower can also layer or wake and create intense tonal noise, with or without
be an important source of noise. In the early development of a self-reinforcing feedback loop [14]. Tonal noise occurs at
wind power, downwind turbines were common and produced low-to-moderate Reynolds numbers (approximately 50,000
high levels of noise associated with the interaction of the tower to 250,000), hence is not usually a problem for large wind
wake with the rotor blades. This form of noise is generated in turbines that operate at higher Reynolds numbers. Small wind
a similar way to the leading edge interaction with turbulent turbines (≤10 kW) may operate at conditions where tonal noise
eddies, though in this case, the eddies are created by the constitutes a major part of the noise source energy. A summary
tower itself. Modern horizontal axis wind turbines place the of the wind turbine noise sources discussed here is given in
rotor upstream of the tower, thus eliminating the wake-rotor Table 1.
interaction. However, the blades still pass through a region of
perturbed flow upstream of the tower [3], creating unsteady lift Table 1. Summary of wind turbine noise sources
and hence noise. Type Directivity Mechanism
Leading-edge Dipole Atmospheric turbulance
interaction noise impinging on rotor trailing edge

Trailing edge Cardioid Boundary layer turbulance

noise passing over rotor trailing edge

Blade tower Dipole Rotor blade passing through


interaction flow perturbed by tower
Tip noise Cardioid Turbulance interacting with
rotor tip
Trailing edge

Airfoil tonal Cardioid Vortex shedding and/or


noise resonant feedback loop on rotor
blade boundary layer



This section will discuss the frequency and time scales
associated with the major aerodynamic noise sources on
Figure 1. The flow over a wind turbine blade tip a horizontal axis wind turbine. These are broadband noise
associated with turbulence leading-edge interaction, airfoil
trailing edge noise and impulsive noise associated with the
Directivity blade-tower interaction. To perform the analyses, the wind
Angle about
trailing edge
turbine used by Oerlemans and Schepers [11] was used. This
turbine is a GE 2.3 MW prototype test turbine with a rotor
diameter of 94 m and a tower height of 100 m. For a wind
speed of 9.75 m/s and a rotational speed of 14.7 RPM, an
empirical model [15] was used to estimate the boundary layer
height at the trailing edge (needed to estimate trailing edge
Blade section noise frequencies). Assuming a tip chord of 1.5 m, the trailing
edge boundary layer height was estimated to be 24 mm at the
tip of the blade (maximum radius).
Broadband Energy
Broadband energy is created by the interaction of turbulence
with the leading and trailing edges. Turbulence leading-edge
interaction noise is dominated by the spectrum of the inflow
Figure 2. Trailing edge noise directivity (high frequency case) turbulence in the atmospheric boundary layer. The peak energy
[3] for this type of noise is contained at a frequency
There are two other, important noise sources that should
be mentioned in this brief review. The first is airfoil tip StVtip
noise, which is generated by flow over the blade tip resulting fpeak = (1)
in a trailing vortex system (see Fig. 1). This form of noise h - 0.7R

8 - Vol. 40, No. 1, April 2012 Acoustics Australia

where the Strouhal number is St = 16.6, h is hub height, Vtip
is the rotor tip speed and R is the blade radius. Using the wind LDT D
= 2πα T (3)
turbine of Oerlemans and Schepers [11], it can be expected that Vtipqcl Vtip
peak energy will occur at approximately 18 Hz. where L is the time derivative of Lift, DT is the tower diameter,
Airfoil trailing edge noise is directly related to the surface q is the dynamic pressure of the flow approaching the blade tip,
pressure spectrum at the trailing edge [8]. There are many well- c is the blade chord, l is the span wise region of the blade under
known empirical models that allow an estimate of the spectral analysis (assumed to be the outer 20% of the rotor blade) and α
energy distribution beneath the airfoil boundary layer. A recent is the time derivative of the blade angle of attack.
and well-validated model is the one by Goody [16]. Using this
model, we are able to estimate the frequency at which most
of the turbulent energy in the boundary layer is converted to Wind
fluctuating surface pressure and hence far-field noise.
Goody [16] shows that surface pressure spectra under
boundary layers can be scaled using the boundary layer height
and that the peak energy is contained approximately a decade
either side of a frequency given by the following relationship
Rotor blade Streamlines about
~1 (2) passing in front of tower
where ω = 2πf , f is frequency, δ is boundary layer height at the
trailing edge and Ue is the velocity external to the boundary
layer at the trailing edge. Using Eq. (2), the trailing edge noise
generated by the blades is expected to have most energy centred
at about 465 Hz. This is in broad agreement with the time-
averaged noise measurements of Oerlemans and Schepers [11],
which show most acoustic energy from the trailing edge of a
wind turbine occurs within the 160-1500 Hz frequency range.
Below 160 Hz, it is expected that the effects of trailing edge Tower
noise will diminish and the effects of turbulence leading edge
noise to become more important.
Blade-Tower Interaction
Impulsive noise may be generated by the interaction of the
blades with the perturbed flow upstream of the tower. Figure 3
Figure 3. Blade tower interaction
illustrates the phenomenon. The flow over the tower creates a
region of non-uniform flow upstream of the tower, represented
Using the turbine described previously, an understanding
by the curved streamlines in Fig. 3. As the rotor blade passes
of the time and frequency scales associated with the BTI can
through this perturbed flow region, the angle of attack changes
be determined. Figure 4 shows the variation of the strength
on the blade, causing a fluctuation in lift force. This fluctuation
of the BTI noise source during one complete revolution of
in lift force creates radiated sound with a time scale associated
the turbine. Time is shown in a non-dimensional form using
with the size of the perturbed flow region upstream of the
the tower diameter and tip speed to determine an appropriate
normalising time scale. The noise source calculation assumes
To estimate the time scales associated with blade-tower
the diameter of the tower DT = 4 m and the rotor disc is
interaction (BTI) a first-order model was created. The model
positioned 4 m upstream of the tower. The calculation was also
uses potential flow theory to estimate the flow field upstream
performed for the blade tip region of the rotor.
of the tower. This is a valid use of potential flow theory as no
As shown in Fig. 4, three pulses are generated during each
boundary layer separation occurs in this region and inviscid
revolution. The creation of each pulse occurs when a blade
effects dominate the flow. Using the flow field estimate, the
passes the tower and interacts with the perturbed flow region.
variation of angle of attack with time is estimated for a blade
Such a repetitive impulsive noise source will contain a variety
section passing though the perturbed flow region. This angle
of frequency components. The autospectrum of the impulsive
of attack history is then converted into a transient lift data
BTI noise source signal is shown in Fig. 5. The spectrum is
record using thin airfoil theory. Using the theory of Curle [17]
shown in non-dimensional units on both axes. The spectral
and assuming a compact source, the source strength can be
decomposition of the BTI noise shows multiple frequency
estimated by taking the time derivative of the lift. Using this
components. The most energy is contained at fDT/Vtip = 0.12 or
method, a first-order estimate of BTI noise source strength,
appropriately non-dimensionalised, is 2.2 Hz and multiple components from fDT/Vtip = 0.04 (0.8 Hz)
to fDT/Vtip ~ 0.6 (11 Hz).

Acoustics Australia Vol. 40, No. 1, April 2012 - 9

0.8 The analysis above also shows that a low frequency noise
source is also present due to the BTI and turbulence leading-
0.6 edge interaction mechanisms. However, the analysis is only
sufficient to predict the dominant frequencies. Determination
0.4 of the strength of these noise sources will depend on many
factors that include the aerodynamic coupling of the blade
2π α̇D T /V tip

and tower, viscous effects on the blade, the dimensions of the

turbine and tower as well as the aeroelastic properties of the
rotor and atmospheric turbulence levels. The analysis provides
assistance to those taking noise measurements and in the
interpretation of existing data.
−0.4 Some observations may be explained by the proposed
models described here. Recent measurements and observations
−0.6 taken at a European wind farm [12] show a marked difference
between day and night. During a summer day, the level of noise
−0.8 from the wind farm was low or not perceivable, even in strong
0 20 40 60 80
tV tip/D T winds (on the ground). On “quiet nights”, residents at distances
of 500-1000 m from the wind farm observed “pile-driving”
Figure 4. Time variation of BTI noise source strength over one noise at a rate coinciding with the blade passing frequency. An
revolution of the GE prototype wind turbine
observer at 1900 m described the noise as an “endless train”.
Within the wind farm (close to the turbines) audible swish-like
noise was observed day and night however, no thumping or
pile-driving noise was audible.
2.5 To explain some of these observations, Van den Berg [12]
pointed out that the state of the atmosphere at night is different
2 to that in the day. In fact, when the atmosphere becomes
stable at night the wind at ground level (and at 10 m which
(V tip/f D T )

is the reference height used to characterise the atmospheric

1.5 boundary layer) can be relatively low while at hub height, it
can be very high. In fact, the hub height wind speed was shown
1 to be 2.6 times higher at night than what would be expected
if the standard day-time atmospheric model was used. This
created 15 dB more noise from the turbine than would be
0.5 expected for the same wind speed at 10m height during the
day. As the ground level wind speed is small, there are low
0 levels of background noise as well thus enhancing the ability
0 0.5 1 1.5 2 of an observer to perceive noise. As wind turbines grow in
f D T /V tip
capacity, this effect can be expected to become greater due to
Figure 5. Autospectrum of the BTI noise source signal the required increase in tower height to accommodate large
radius rotors.
Using A-weighted noise measurements taken over a 50 ms
WIND TURBINE NOISE MEASUREMENTS time-base, Van den Berg [12] was able to show that the noise
level fluctuated at a rate of about 1 Hz at a residence’s home
The preceding analysis gives an indication of the frequency
750 m from the wind farm. The amplitude of this fluctuation
scales that we can expect from three dominant wind turbine
varied between 1 and 5 dB at various times throughout the
aerodynamic noise sources. Note that there are more possible
measurement period. It was inferred that this variation was due
sources and these may also have significant contribution to the
to periods of time when noise emission from multiple wind
observed noise, but this paper will concentrate on blade swish
turbines in the farm become in or out of phase. Van den Berg
and BTI to explain observed behaviour.
[12] states that this is the cause of the impulsive noise observed
Broadband noise at relatively high frequency is the
outside of the wind farm. Residents expressed that the noise
dominant component of blade swish. Although modulated
is more annoying at night when the rotor speed is high, thus
at the blade passing frequency (~1 Hz), blade swish cannot
linking the stability of the atmosphere to annoyance.
be considered a low frequency noise source. Rather, it is an
The analysis of the previous section is now used to
amplitude modulated broadband source with dominant energy
explain these observations. The time varying measurements
at about 500 Hz (for the example turbine in this paper). Swish
are A-weighted and therefore are dominated by noise with
has been recorded from wind turbines for many years [11, 18]
frequencies that are linked to trailing edge noise. The amplitude
and can be attributed to noise generated at the trailing edge of
modulation observed is hence not due to the interaction with
the outer part of the turbine and its forward looking directivity
the tower but is due to the unique directivity associated with
pattern coupled with blade rotation.

10 - Vol. 40, No. 1, April 2012 Acoustics Australia

the trailing edge source. The reinforcement effects observed by BTI-excited structural vibration may also be apparent inside
Van den Berg [12] are still caused by multiple turbines except the home. While much more work is required to understand
that the sound is emitted directly from the trailing edge rather BTI and swish reinforcement, the model presented provides a
than from BTI, as suggested by Van den Berg in Ref. [19]. framework for understanding and addressing public concerns
This is not to suggest that the BTI source is not important. about wind turbine noise.
In the same way as the broadband swish noise can be reinforced
Preliminary wind turbine measurements in South Australia
and become unexpectedly high outside of a wind farm, it is
not unreasonable to expect that the same may be true for BTI To investigate amplitude modulation of operational wind
noise. Currently, there is no methodology or dataset available turbines, a series of daytime measurements were taken at a South
that can allow researchers to accurately quantify BTI noise. Australian wind farm. Acoustic data from a line of 7 turbines
However, high levels of low-frequency BTI noise may couple (6 of which were operative) were recorded on a November
with structural resonances of homes and workplaces, creating afternoon in 2011 at a sampling frequency of 51.2 kHz using a
audible noise that may have an annoying character. As wind Brüel and Kjær 4190 ½ inch free-field microphone connected to
turbines become larger, the BTI noise source can be expected a National Instruments Data Acquisition system (NIDAQ 9234).
to become stronger. A similar argument may be applicable to The microphone was located at broadside to the wind farm
turbulence leading-edge interaction noise as well, albeit with at a distance of several hundred metres. The microphone was
dominant energy levels at higher frequencies. covered in a foam windsock, was held in a microphone stand at
0.75 m height and was directed towards the nearest turbine.
Conditions were sunny with very little cloud cover. The
BTI Reinforcement Swish Reinforcement microphone was located downwind of the wind farm. Wind
speed was not measured directly; however, a wind speed
of 17 km/h (4.72 m/s) was recorded at the closest Bureau
Wind of Meteorology weather station on the afternoon of the
measurements. Noise from the wind farm was clearly audible.
Acoustic data were bandpass filtered to 500-5000 Hz. A
12-second long time series of the measured data is shown in
Fig. 7. The signal amplitude is observed to fluctuate temporally,
with elements of periodicity apparent.
The A-weighted sound pressure level (SPL) of the signal is
shown in Fig. 8, which was calculated by separating the signal into
125 ms long segments, performing a fast Fourier transform on
each segment then applying an A-weighted filter and integrating
to obtain a mean energy (equivalent to the time weighting FAST
Wind Turbine 1 Wind Turbine 2 setting on a sound level meter). The single SPL value from each
time segment was then plotted in Fig. 8, yielding an A-weighted
Figure 6. Plan view of two wind turbines with possible zones of noise SPL as a function of time. It can be seen that the periodicity in
reinforcement the signal amplitude becomes more apparent and these periodic
amplitude fluctuations are observed to dominate the signal. The
The reinforcement of trailing edge and BTI noise sources expected signal maxima and minima corresponding to a 1.28
may create regions about the wind farm where noise fluctuation second period are shown in the figure, and although not every
amplitudes are high. As a means to explain wind farm noise point corresponds to a maxima or minima, the trend is apparent.
reinforcement, a simple schematic showing two wind turbines The 1.28 second period is within 1.2% of the wind turbine blade
in plan view is displayed in Fig. 6. It shows noise propagating pass frequency estimated from video footage, supporting the
upwind of the turbines only (other directions are omitted for hypothesis that the amplitude fluctuations are due to amplitude
clarity) and regions where broadband swish noise and BTI modulation at the blade passing frequency. Figure 8 also shows,
noise may be reinforced. Of course, the sound will couple for comparison, acoustic data recorded on the same afternoon
with atmospheric propagation effects making the actual sound (and using the same methodology) at a location further from the
paths more complicated than is represented in the figure, wind turbine farm where wind turbine noise was not audible. By
but conceptually the idea is the same. Note that BTI noise comparing the two data sets, it is apparent that both the amplitude
signals, as described in this paper, may cancel each other as of noise within the 500-5000 Hz range, and the amplitude of
well reinforce as they are pulses of temporally coherent sound; any temporal fluctuations are significantly smaller in this second
however, the broadband noise signals are incoherent random measurement.
signals and may only reinforce and not cancel each other. If this
model is correct, it may explain why some residents become
annoyed, both inside and outside a home. While broadband
swish noise may annoy people outside, its high frequency
components may be attenuated inside a home. However, if
BTI reinforcement occurs at the same location, noise from

Acoustics Australia Vol. 40, No. 1, April 2012 - 11

airfoil designs. One design achieved a 2.9 dB OASPL noise
reduction (over the original NACA 0012 shape used to start
the optimisation process) whilst also reducing drag. It can be
expected that much quieter airfoil designs will be developed as
noise prediction methods become more accurate and efficient.
Another important passive noise control technique for
trailing edge noise is the use of trailing edge serrations. These
are saw-tooth extensions placed on the trailing edge. As
originally pointed out by Howe [23], the serrations present a
trailing edge at an angle to the stream wise flow direction thus
reducing the efficiency of the edge sound source. Theoretically,
serrations are able to reduce noise by a large amount. However,
in practice, serrations do not reduce noise as much as theory
suggests [9, 24, 25] and this may be due to the production
or re-orientation of turbulence by the serrations themselves.
Porous trailing edge inserts [26] are also promising noise
reducing devices, but may have limited applicability due
to dirt accumulation in the pores, requiring regular costly
Figure 7. Time series of acoustic data
While shape modifications or inserts may provide an
effective means of trailing edge (broadband swish) noise
control, passive means of BTI noise are limited. One answer
is to increase the distance between the rotor tip and tower. The
current spacing between the rotor tip and tower has probably
been maximised by the manufacturer. Increasing this distance
will require extensive redesign of the gearbox and nacelle and
could introduce more problems such as shortened mechanical
life, vibration and noise.
Active Control Concepts
Swish and BTI reinforcement occurs due to in-phase
noise production on multiple wind turbines. As each turbine
rotates in the same direction and experiences close to the same
wind speed and direction they will turn at very nearly the
same angular velocity. If the azimuthal phase of a group of
wind turbines is nearly the same, then we would expect that
their sound would be produced at nearly the same time and
Figure 8. A-weighted SPL with 125 ms FFTs (FAST): near-turbine
propagate in a similar manner. Given that broadband swish
measurements (black) and measurements (not simultaneous) at a
nearby location where wind turbines were not audible (grey). Circles has a forward propagating directivity, then zones of high
and triangles with dashed lines represent expected signal maxima and amplitude modulation of trailing edge noise are expected. BTI
minima corresponding to blade pass frequency, respectively noise has the directivity of a dipole, hence an array of in-phase
BTI sources will create alternate zones of reinforcement and
WIND TURBINE NOISE CONTROL Active phase desynchronisation is a concept that can
CONCEPTS potentially alleviate this situation. By monitoring the phase of
This section of the paper will outline methods of controlling each blade in a wind farm, small adjustments to the rotor blade
both broadband swish and BTI noise. pitch or brake can be made to alter the blade’s phase and ensure
that noise reinforcement does not occur at a particular receiver
Passive Control Methods location or locations, such as homes. While this seems a simple
The most efficient means of controlling trailing edge noise and cost effective solution to the problem, it may be difficult to
is to reduce the strength of its source. One of the most direct implement without more knowledge of how the noise sources
methods for doing this is to alter the blade shape in order to are produced, their strengths and how they propagate in the
influence the nature of the turbulent boundary layer at the atmosphere.
trailing edge. Methods of doing this vary between ad-hoc
design changes to computationally demanding aeroacoustic
shape optimisation [20, 21]. Recently, Jones et al. [22]
developed an optimisation procedure using a semi-empirical
model of trailing edge noise to develop new, low noise

12 - Vol. 40, No. 1, April 2012 Acoustics Australia

SUMMARY AND OUTLOOK [9] S. Oerlemans, M. Fisher, T. Maeder, and K. Kögler, “Reduction
of wind turbine noise using optimized airfoils and trailing-edge
This paper has reviewed the major sources of aerodynamic
serrations”, AIAA Journal 47(6), 1470-1481 (2009)
noise on modern horizontal wind turbines. A brief analysis of [10] P. Migliore and S. Oerlemans, “Wind tunnel aeroacoustic tests
the time and frequency scales of two dominant noise sources of six airfoils for use on small wind turbines”, Journal of Solar
for a modern wind turbine was presented. Broadband airfoil Energy Engineering 126, 974-985 (2004)
trailing edge noise for the case studied was shown to have most [11] S. Oerlemans and J.G. Schepers, “Prediction of wind turbine
of its energy at approximately 500 Hz. Its directivity ensures noise and validation against experiment”, International Journal
that trailing edge noise from a wind turbine will have its of Aeroacoustics 8(6), 555–584 (2009)
amplitude modulated with time at the blade passing frequency. [12] G.P. Van den Berg, “Effects of the wind profile at night on wind
While the amplitude modulation occurs at low frequency, it turbine sound”, Journal of Sound and Vibration 277, 955–970
cannot be considered a low frequency noise source. Blade- (2004)
[13] E. Arcondoulis, C.J. Doolan, A. Zander, and L.A. Brooks,
tower interaction (BTI) noise was analysed using a first order
“A review of trailing edge noise generated by airfoils at low
model and its frequency content was found to have maximum
to moderate Reynolds number”, Acoustics Australia 38(3),
energy at 2.2 Hz. 129–133 (2010)
Some measurements from a modern European wind farm [14] D.J. Moreau, L.A. Brooks and C.J. Doolan, “On the aeroacoustic
were reviewed. These results strongly suggest that noise from tonal noise generation mechanism of a sharp-edged plate”,
multiple wind turbines in a wind farm can reinforce each other Journal of the Acoustical Society of America 129(4), EL154-
and create impulsive “pile-driving” like sound, considerable EL160, 2011.
distances from the wind farm. The published results are [15] T. Cebeci and P. Bradshaw, Momentum transfer in boundary
A-weighted; hence are dominated by noise from the broadband layers, McGraw Hill, 1977
swish (trailing edge) component. Recent measurements taken [16] M. Goody, “Empirical spectral model of surface pressure
close to a South Australian wind farm confirm that amplitude fluctuations”, AIAA Journal 42(9), 1788–1794 (2004)
[17] N. Curle, “The influence of solid boundaries on aerodynamic
modulation is present under Australian daytime conditions. It
sound”, Proceedings of the Royal Society of London A, 231,
is speculated BTI noise may also be reinforced in the same
505–514 (1955)
manner and create zones of high-level low-frequency sound. [18] H.H. Hubbard, F.W. Grosveld and K.P. Shepherd, “Noise
Passive and active control concepts were presented with active characteristics of large wind turbine generators”, Noise Control
phase desynchronisation a promising method for controlling Engineering Journal 21, 21-29 (1983)
both forms of noise. [19] G.P. Van Den Berg, “The beat is getting stronger: the effect of
More research is needed to understand both swish and BTI atmospheric stability on low frequency modulated sound of
noise sources before effective control methods can be pursued. wind turbines”, Noise Notes 4, 15–40 (2005)
BTI noise remains the least well studied and some controversy [20] A.L. Marsden, M. Wang, J.E Dennis and P. Moin, “Trailing-
surrounds the issue of whether it is a significant noise source. edge noise reduction using derivative-free optimization and
Only more detailed measurements and understanding of how it large-eddy simulation”, Journal of Fluid Mechanics 572,
13–36 (2007)
is generated and propagates will provide meaningful answers.
[21] T. Lutz, A. Herrig, W. Wurz, M. Kamruzzaman and E. Kramer,
“Design and wind-tunnel verification of low-noise airfoils for
REFERENCES wind turbines”, AIAA Journal 45(4), 779-785 (2007)
[1] Geoscience-Australia and ABARE, Australian energy resource [22] R.F. Jones, C.J. Doolan and M. Teubner, “Minimization of
assessment, Technical report, Australian Government, trailing edge noise by parametric airfoil shape modifications”,
Canberra, Australia, 2010. Proceedings of the 17th AIAA/CEAS Aeroacoustics Conference
[2] E. Pedersen and K.P. Waye, “Perception and annoyance (32nd AIAA Aeroacoustics Conference), AIAA 2011-2782,
due to wind turbine noise—a dose–response relationship”, Portland, Oregon, USA, 6-8 June 2011
Journal of the Acoustical Society of America 116(6), 3460- [23] M.S. Howe, “Aerodynamic noise of a serrated trailing edge”,
3470 (2004) Journal of Fluids and Structures 5(1), 33–45 (1991)
[3] S. Wagner, R. Bareiss and G. Guidati, Wind turbine noise, [24] D.J. Moreau, L.A. Brooks and C.J. Doolan, “Flat plate self-
Springer Verlag, 1996 noise reduction at low-to-moderate Reynolds number with
[4] S. Oerlemans, P. Sijtsma and B. Mendez Lopez, “Location and trailing edge serrations”, Proceedings of Acoustics 2011, Gold
quantification of noise sources on a wind turbine”, Journal of Coast, Australia, 2-4 November 2011
Sound and Vibration 299, 869–883 (2007) [25] M. Gruber, P. Joseph, and T.P. Chong, “Experimental
[5] W.K. Blake, Mechanics of flow-induced sound and vibration, investigation of airfoil self noise and turbulent wake reduction
Vol. 2, Academic Press, New York, 1986 by the use of trailing edge serrations”, Proceedings of the
[6] M.J. Lighthill, “On sound generated aerodynamically: General 16th AIAA/CEAS Aeroacoustics Conference (31st AIAA
theory”, Proceedings of the Royal Society of London A 221, Aeroacoustics Conference), AIAA 2010-3803, Stockholm,
564–587 (1952) Sweden, 7-9 June 2010
[7] J.E. Ffowcs-Williams and L.H. Hall, “Aerodynamic sound [26] T. Geyer, E. Sarradj and C. Fritzsche, “Measurement of
generation by turbulent flow in the vicinity of a scattering the noise generation at the trailing edge of porous airfoils”,
half plane”, Journal of Fluid Mechanics 40(4), 657–670 Experiments in Fluids 48(2), 291–308 (2010)
[8] M.S. Howe, “A review of the theory of trailing edge noise”,
Journal of Sound and Vibration 61(3), 437-465, (1978)

Acoustics Australia Vol. 40, No. 1, April 2012 - 13